A manufacturing method for pulsed high-voltage silicon stack

Abstract

This application relates to the field of semiconductor manufacturing technology, and particularly to a method for manufacturing a pulsed high-voltage silicon stack. The method includes: analyzing the change trends of the simulation results of various optimization targets under any type of circuit parameter to be optimized in the current iteration and all previous iterations to determine the comprehensive simulation change coefficient in the current iteration; analyzing the distances between the position update coordinates of each rime particle in each iteration and its adjacent previous iteration, and evaluating the change trends of the position change amounts in the current iteration and all previous iterations to determine the position change degree of each rime particle in the current iteration, and combining the comprehensive simulation change coefficient to optimize the parameters to be optimized of the pulsed high-voltage silicon stack. This application improves the accuracy of the rime particles converging to the pareto optimal solution and the performance of the pulsed high-voltage silicon stack product by dynamically adjusting the adhesion of the rime particles.

Description

Technical Field

[0001] The present application relates to the field of semiconductor manufacturing technology, and in particular to a method for manufacturing a pulsed high-voltage silicon stack. Background Art

[0002] The pulsed high-voltage silicon stack is a new type of special semiconductor device, usually composed of multiple silicon pulsed high-voltage diode cores packaged in series and parallel, with an operating voltage between tens of thousands of volts and hundreds of thousands of volts, and a pulse current between thousands of amperes and hundreds of thousands of amperes. It is often used in the field of electromagnetic catapults in large-scale scientific research experimental equipment related to nuclear physics, synchrotron radiation light sources or other large pulse power modulators for pulse isolation, reverse suppression protection or modulation waveforms. In the manufacturing process of the pulsed high-voltage silicon stack, in order to reduce production costs, simulation software is usually used to simulate and analyze the circuit parameters of the pulsed high-voltage silicon stack, and based on the simulation results, the circuit parameters of the pulsed high-voltage silicon stack are optimized using a multi-objective optimization algorithm to improve the performance of the pulsed high-voltage silicon stack, such as improving the uniformity of voltage distribution and voltage utilization in the pulsed high-voltage silicon stack.

[0003] The Multi-objective rime optimization algorithm (MORIME) is a multi-objective extension of the rime optimization algorithm (RIME). It is a new type of heuristic optimization algorithm. On the classic IEEECEC2022 test set, the RIME algorithm has superior performance compared to some intelligent optimization algorithms. However, when solving complex multi-objective optimization problems, such as circuit parameter optimization problems with highly nonlinear and complex interactions such as pulsed high-voltage silicon stacks, the growth environment or attached object model of rime in the MORIME algorithm will change. Therefore, it is necessary to modify the parameters in the MORIME algorithm to avoid falling into the local optimal solution when directly using the MORIME algorithm to optimize the circuit parameters of the pulsed high-voltage silicon stack, which will affect the selection of the best circuit parameters in the pulsed high-voltage silicon stack, thereby reducing the performance of the pulsed high-voltage silicon stack product. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a method for manufacturing a pulsed high voltage silicon stack to solve the existing problems.

[0005] A pulsed high-voltage silicon stack manufacturing method of the present application adopts the following technical solution:

[0006] An embodiment of the present application provides a method for manufacturing a pulsed high voltage silicon stack, the method comprising the following steps:

[0007] Obtain various circuit parameters and targets to be optimized of the pulsed high-voltage silicon stack, and use the multi-objective rime algorithm to iteratively optimize the various circuit parameters to be optimized. The specific process is as follows:

[0008] S1: For the current iteration and all previous iterations, obtain the simulation results of various optimization targets of the pulsed high-voltage silicon stack under different circuit parameters to be optimized for each iteration, and the updated position coordinates of each rime particle at the end of each iteration.

[0009] S2: By analyzing the change trends of the simulation results of various optimization targets under any type of circuit parameter to be optimized for the current iteration and all previous iterations, determine the simulation change value of various optimization targets in the pulsed high-voltage silicon stack under any type of parameter to be optimized for the current iteration. Based on the average distribution of the simulation change values of various optimization targets under all types of parameters to be optimized, determine the simulation change coefficient of various optimization targets for the current iteration to determine the comprehensive simulation change coefficient for the current iteration.

[0010] S3: By analyzing the distances between the updated position coordinates of each rime particle for each iteration and its adjacent previous iteration, determine the position change amount of each rime particle for each iteration; evaluate the change trends of the position change amounts for the current iteration and all previous iterations, determine the position change trend value of each rime particle for the current iteration, and combine with the position change amount for the current iteration to determine the position change degree of each rime particle for the current iteration.

[0011] S4: Based on the comprehensive simulation change coefficient and the position change degree, determine the adhesion adjustment value of each rime particle for the current iteration to optimize the parameters to be optimized of the pulsed high-voltage silicon stack.

[0012] Preferably, the obtaining of various circuit parameters to be optimized and various optimization targets of the pulsed high-voltage silicon stack includes: all types of circuit parameters to be optimized at least include reverse internal resistance, ground capacitance of the silicon rectifier tube, distributed capacitance to the high-voltage end, number of series tubes, and PN junction reverse barrier capacitance; all types of optimization targets at least include voltage uniformity coefficient, voltage utilization rate, and reverse breakdown voltage value.

[0013] Preferably, the method for determining the simulation change value of various optimization targets in the pulsed high-voltage silicon stack under any type of parameter to be optimized for the current iteration is:

[0014] For the current iteration and all previous iterations, fit the simulation results of various optimization targets under any type of circuit parameter to be optimized to obtain a fitting line, and use the slope of the fitting line as the simulation change value of various optimization targets in the pulsed high-voltage silicon stack under any type of circuit parameter to be optimized for the current iteration.

[0015] Preferably, the simulation change coefficient of each optimization target under the current iteration is the mean value of the simulation change coefficients of each optimization target in the pulse high-voltage silicon stack under all types of circuit parameters to be optimized under the current iteration.

[0016] Preferably, the expression of the comprehensive simulation change coefficient under the current iteration is: ; where represents the comprehensive simulation change coefficient under the current iteration; represents the number of simulation change coefficients greater than 0 among the simulation change coefficients of all optimization targets under the current iteration and all previous iterations; represents the simulation change coefficient of the nth type of optimization target in the pulse high-voltage silicon stack under the current iteration; N represents the total number of types of optimization targets in the high-voltage silicon stack under the current iteration; exp( ) represents the exponential function with the natural constant as the base.

[0017] Preferably, the position change amount of each rime particle under each iteration is the Euclidean distance between the position update coordinates of each rime particle under each iteration and its adjacent previous iteration.

[0018] Preferably, the method for determining the position change trend value of each rime particle under the current iteration is:

[0019] Taking the position change amounts of each rime particle under the current iteration and all previous iterations as the ordinate for fitting, obtaining the fitted straight line and denoting it as the position fitting straight line, and taking the slope of the position fitting straight line as the position change trend value of each rime particle under the current iteration.

[0020] Preferably, the expression of the position change degree of each rime particle under the current iteration is: ; where represents the position change degree of the ath rime particle under the current iteration; represents the position change amount of the ath rime particle under the current iteration; represents the position change trend value of the ath rime particle under the current iteration; exp( ) represents the exponential function with the natural constant as the base.

[0021] Preferably, the expression of the adhesion adjustment value of each rime particle under the current iteration is: ; where represents the adhesion adjustment value of the ath rime particle under the current iteration; S represents the comprehensive simulation change coefficient under the current iteration; represents the position change degree of the ath rime particle under the current iteration; norm( ) represents the normalization function; represents a preset constant greater than 0.

[0022] Preferably, optimizing the parameters to be optimized of the pulsed high-voltage silicon stack includes:

[0023] Taking all types of circuit parameters to be optimized in the pulsed high-voltage silicon stack as the input of the multi-objective rime algorithm. Among them, the adhesion adjustment value of each rime particle in the current iteration is used as the adhesion of the corresponding rime particle in the next iteration of the multi-objective rime algorithm until the iteration is completed to obtain all types of optimal circuit parameters in the pulsed high-voltage silicon stack.

[0024] This application has at least the following beneficial effects:

[0025] By analyzing the change trends of the simulation results of various optimization targets of the pulsed high-voltage silicon stack under different circuit parameters to be optimized, this application constructs a comprehensive simulation change coefficient, which can effectively evaluate the change of the simulation results of multiple optimization targets of the pulsed high-voltage silicon stack, thereby detecting and analyzing the change trend of the objective function, judging whether the optimization process is developing in the correct direction, and thus timely and accurately adjusting the optimization strategy to avoid the situation that the multi-objective rime algorithm falls into a local optimal solution; further, by analyzing the change of the position coordinates of the rime particles under different iterations, a position change degree is constructed, which can guide the optimization algorithm to adjust the adhesion of the rime particles during the iteration process, improve the accuracy of the algorithm converging to the optimal solution, and thus improve the performance of the pulsed high-voltage silicon stack product; by combining the position change degree and the comprehensive simulation change coefficient, this application dynamically adjusts the adhesion of the rime particles, thereby effectively avoiding the situation that the optimization algorithm falls into a local optimal solution, improving the accuracy of the rime particles converging to the pareto optimal solution, and further obtaining better circuit parameters of the pulsed high-voltage silicon stack, improving the performance of the pulsed high-voltage silicon stack product. Description of the Drawings

[0026] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0027] Figure 1 It is a flowchart of the steps of a method for manufacturing a pulsed high-voltage silicon stack provided by an embodiment of the present application;

[0028] Figure 2 It is a schematic diagram of the extraction process of the adhesion adjustment value provided by an embodiment of the present application. Detailed Embodiments

[0029] In order to further elaborate on the technical means and effects adopted by this application to achieve the intended invention purpose, the following will, in conjunction with the accompanying drawings and preferred embodiments, elaborate in detail on a method for manufacturing a pulsed high-voltage silicon stack proposed according to this application, its specific implementation manner, structure, features, and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs.

[0031] The following will specifically describe the specific solution of a method for manufacturing a pulsed high-voltage silicon stack provided by this application in conjunction with the accompanying drawings.

[0032] A method for manufacturing a pulsed high-voltage silicon stack provided by an embodiment of this application, specifically, provides the following method for manufacturing a pulsed high-voltage silicon stack. Please refer to Figure 1 , and the method includes the following steps:

[0033] S1: For the current iteration and the preset number of iterations before it, obtain the simulation results of various optimization targets of the pulsed high-voltage silicon stack under different circuit parameters to be optimized for each iteration, and the updated coordinates of the positions of each rime particle at the end of each iteration.

[0034] According to the equivalent circuit during the reverse operation of the high-voltage silicon stack, use ATP software to build a simulation model of the pulsed high-voltage silicon stack. In this embodiment, the reverse internal resistance of the silicon rectifier tube, the capacitance of the silicon rectifier tube to the ground, the distributed capacitance to the high-voltage end, the number of series tubes (the total number of silicon rectifier tubes in the simulation model), and the reverse barrier capacitance of the PN junction in the simulation model of the pulsed high-voltage silicon stack are used as the circuit parameters to be optimized for the pulsed high-voltage silicon stack; in this embodiment, the voltage uniformity coefficient, voltage utilization rate, and reverse breakdown voltage value of the pulsed high-voltage silicon stack are used as the optimization targets for the pulsed high-voltage silicon stack.

[0035] Furthermore, use the multi-objective rime algorithm to iteratively optimize various circuit parameters to be optimized. For the current iteration and the preset number of iterations before it, obtain the simulation results of various optimization targets of the pulsed high-voltage silicon stack under different circuit parameters to be optimized for each iteration, and the updated coordinates of the positions of each rime particle at the end of each iteration.

[0036] S2: By analyzing the variation trends of the simulation results of various optimization targets under any type of circuit parameter to be optimized in the current iteration and all previous iterations, determine the simulation change values of various optimization targets in the pulsed high-voltage silicon stack under any type of parameter to be optimized in the current iteration. Based on the average distribution of the simulation change values of various optimization targets under all types of parameters to be optimized, determine the simulation change coefficients of various optimization targets in the current iteration, so as to determine the comprehensive simulation change coefficient in the current iteration.

[0037] In the multi-objective rime algorithm, the adhesion degree of rime particles is used to control the distance between the centers of two rime particles, where the adhesion degree is a random number between 0 and 1. Different circuit parameters of the pulsed high-voltage silicon stack will have different effects on the optimization targets of the pulsed high-voltage silicon stack. For example, increasing the reverse barrier capacitance of the PN junction will reduce the voltage uniformity coefficient, decreasing the ground capacitance of the silicon rectifier will increase the voltage uniformity coefficient, and increasing the number of series-connected tubes of the silicon rectifier will increase the voltage uniformity coefficient while reducing the voltage utilization rate, etc. When directly using the multi-objective rime algorithm to optimize the circuit parameters of the pulsed high-voltage silicon stack, the randomness of the adhesion degree in the algorithm will cause the rime particles to easily converge too quickly to a local search space that simultaneously reduces the results of multiple optimization targets of the pulsed high-voltage silicon stack during its iterative process, making it difficult for the multi-objective rime algorithm to converge to the pareto optimal front in a limited number of iterations, and thus unable to provide a set of optimal circuit parameters that balance all the optimization targets of the pulsed high-voltage silicon stack.

[0038] Therefore, during the iterative optimization of the parameters to be optimized using the multi-objective rime algorithm, if the simulation results of multiple optimization targets of the pulsed high-voltage silicon stack show a large decreasing trend, it indicates that the rime particles may have fallen into a local search space that simultaneously reduces the results of multiple optimization targets of the pulsed high-voltage silicon stack, and it is very difficult for the rime particles to jump out of the local search space in subsequent iterations. Therefore, in this embodiment, by increasing the adhesion degree of the rime particles, the search range of the rime particles is expanded, so that the rime particles can jump out of the local search space in subsequent iterations to avoid the subsequent rime particles from converging to the local optimal solution;

[0039] On the contrary, during the iterative process of the rime particles, if the simulation results of multiple optimization targets of the pulsed high-voltage silicon stack show a large increasing trend, it indicates that the rime particles have entered the local search space containing the pareto optimal solution at this time, and may jump out of the local search space in subsequent iterations. Therefore, in this embodiment, by reducing the adhesion degree of the rime particles, the search range of the rime particles is reduced, so that the rime particles can iterate in the search space where the pareto optimal solution is located, to improve the accuracy of the subsequent rime particles converging to the pareto optimal solution.

[0040] Based on the above analysis, first, the result of multiplying all the optimization targets of the pulse high-voltage silicon stack is used as the objective function of the multi-objective rime algorithm. This is because improving the uniformity of the distribution of the reverse voltages on the silicon rectifier tubes in the pulse high-voltage silicon stack and the maximum reverse voltage value that the pulse high-voltage silicon stack can withstand, that is, improving the voltage uniformity coefficient and the reverse breakdown voltage value of the pulse high-voltage silicon stack, can avoid the silicon rectifier tube with the maximum reverse voltage in the pulse high-voltage silicon stack from being broken down; and improving the voltage utilization rate of the pulse high-voltage silicon stack can reduce the energy loss during the use of the pulse high-voltage silicon stack.

[0041] Secondly, all types of circuit parameters to be optimized of the pulse high-voltage silicon stack are respectively used as each dimension in the search space of the multi-objective rime algorithm, that is, each dimension represents a type of circuit parameter to be optimized, and the upper and lower limit values of each type of circuit parameter to be optimized are respectively used as the upper and lower bounds of the corresponding dimension.

[0042] Among them, in this embodiment, the initial rime population, the maximum number of iterations, and the size of the external archive in the multi-objective rime algorithm are respectively set to 50, 100, and 100. The size of the external archive shall not be less than the size of the initial rime population. The control coefficient of the environmental factor in the multi-objective rime algorithm takes a value of 5. The values of the initial rime population quantity, the maximum number of iterations, the size of the external archive, and the control coefficient of the environmental factor in the multi-objective rime algorithm are all set artificially, and their values are not specially limited in this embodiment. Implementers can also set them according to specific situations by themselves, and this embodiment does not make special restrictions.

[0043] Furthermore, for the current iteration and all previous iterations, the simulation results of all types of optimization targets under any type of circuit parameter to be optimized are fitted to obtain a fitting line, and the slope of the fitting line is used as the simulation change value of all types of optimization targets in the pulse high-voltage silicon stack under any type of circuit parameter to be optimized in the current iteration.

[0044] It should be noted that there are many commonly used fitting algorithms. In this embodiment, the least squares fitting algorithm is used to fit the simulation results of all types of optimization targets under any type of circuit parameter to be optimized for the current iteration and all previous iterations to obtain a fitting line. Implementers can also use other fitting methods such as linear regression. The selection of the fitting method is not specially limited in this embodiment. Among them, all the contents related to fitting in this embodiment adopt the least squares fitting algorithm. The least squares fitting algorithm is a well-known technology, and its specific fitting principle process will not be elaborated here.

[0045] Furthermore, the mean value of the simulation change coefficients of all types of optimization targets in the pulse high-voltage silicon stack under all types of circuit parameters to be optimized in the current iteration is used as the simulation change coefficient of all types of optimization targets in the current iteration.

[0046] Comprehensive simulation change coefficient under the current iteration The expression is as follows: ; In the formula, represents the number of simulation change coefficients greater than 0 among the simulation change coefficients of all the targets to be optimized under the current iteration and all previous iterations; represents the simulation change coefficient of the nth type of target to be optimized in the pulse high-voltage silicon stack under the current iteration; N represents the total number of types of targets to be optimized in the high-voltage silicon stack under the current iteration; exp( ) represents the exponential function with the natural constant as the base.

[0047] It can be understood from the comprehensive simulation change coefficient under the current iteration that during the current iteration and all previous iterations, if the number of simulation change coefficients greater than 0 among the simulation change coefficients of all the targets to be optimized under the current iteration and all previous iterations is larger, that is, the number of targets to be optimized with an increasing trend in the simulation results of the targets to be optimized of the pulse high-voltage silicon stack is larger, and the simulation change coefficients of various types of targets to be optimized in the pulse high-voltage silicon stack under the current iteration are larger, that is, the degree of the overall increasing trend of the simulation results of the targets to be optimized is larger, then the comprehensive simulation change coefficient is larger, indicating that there is a relatively large increasing trend in the simulation results of multiple targets to be optimized of the pulse high-voltage silicon stack during the current iteration;

[0048] On the contrary, if the number of simulation change coefficients greater than 0 among the simulation change coefficients of all the targets to be optimized under the current iteration and all previous iterations is smaller, that is, the number of targets to be optimized with an increasing trend in the simulation results of the targets to be optimized of the pulse high-voltage silicon stack is smaller, and the simulation change coefficients of various types of targets to be optimized in the pulse high-voltage silicon stack under the current iteration are smaller, that is, the degree of the overall increasing trend of the simulation results of the targets to be optimized is smaller, then the comprehensive simulation change coefficient is smaller.

[0049] S3: By analyzing the distance between the position update coordinates of each rime particle under each iteration and its adjacent previous iteration, determine the position change amount of each rime particle under each iteration; evaluate the change trend of the position change amount under the current iteration and all previous iterations, determine the position change trend value of each rime particle under the current iteration, and combine the position change amount under the current iteration to determine the position change degree of each rime particle under the current iteration.

[0050] In the process of iteratively optimizing the parameters to be optimized using the multi-objective rime algorithm, if the positions of the rime particles do not change significantly during the iteration, it indicates that the rime particles may have fallen into a local search space that simultaneously reduces the results of multiple optimization objectives of the pulsed high-voltage silicon stack. It is difficult for the rime particles to jump out of the local search space in subsequent iterations. On the contrary, during the iteration of the rime particles, if the positions of the rime particles change significantly in their subsequent iterations, it indicates that the rime particles have entered the local search space containing the Pareto optimal solution at this time and may jump out of the local search space in subsequent iterations. Therefore, in this embodiment, the adhesion degree of the rime particles is reduced to narrow the search range of the rime particles, so that the rime particles are iterated in the search space where the Pareto optimal solution is located, so as to improve the accuracy of the subsequent rime particles converging to the Pareto optimal solution.

[0051] Based on the above analysis, in this embodiment, by analyzing the position change trend of the rime particles, the adhesion degree of the rime particles is increased or decreased to narrow the search range of the rime particles, so that the rime particles are iterated in the search space where the Pareto optimal solution is located, so as to improve the accuracy of the subsequent rime particles converging to the Pareto optimal solution. Specifically:

[0052] The Euclidean distance between the coordinates of the position update of each rime particle in each iteration and the coordinates of the position update of the immediately preceding iteration is used as the position change amount of each rime particle in each iteration; it is used to characterize the change amount between the position of the rime particle at the end of each iteration and the position at the end of the previous iteration.

[0053] Among them, the calculation method of the Euclidean distance is a well-known technology, and its specific calculation process will not be elaborated here.

[0054] In particular, the method for determining the position change amount of the first iteration is: the Euclidean distance between the position coordinates of the position update of the rime particle in the first iteration and the initial position coordinates of the rime particle is used as the position change amount of the rime particle in the first iteration, where the initial position coordinates of the rime particle can be obtained through the multi-objective rime algorithm.

[0055] Furthermore, the position change amounts of each rime particle in the current iteration and all previous iterations are used as the ordinate for fitting to obtain a fitted line, denoted as the position fitted line, and the slope of the position fitted line is used as the position change trend value of each rime particle in the current iteration.

[0056] The position change degree of the a-th rime particle in the current iteration The expression is: ; in the formula, represents the position change amount of the a-th rime particle in the current iteration; It represents the position change trend value of the $a$-th rime particle in the current iteration; $\exp( )$ represents the exponential function with the natural constant as the base.

[0057] From the degree of position change of each rime particle in the current iteration, it can be understood that if the position change amount of the current rime particle is larger and the position change trend value is larger, the degree of position change of the obtained rime particle is larger, indicating that the position of the rime particle has changed significantly compared to the previous iteration, and it is more necessary to reduce the adhesion of the rime particle to narrow the search range of the rime particle; conversely, if the position change amount of the current rime particle is smaller and the position change trend value is smaller, the degree of position change of the obtained rime particle is smaller, indicating that the position of the rime particle has not changed significantly compared to the previous iteration, and it is more necessary to increase the adhesion of the rime particle to increase the search range of the rime particle.

[0058] S4: Based on the comprehensive simulation change coefficient and the degree of position change, determine the adhesion adjustment value of each rime particle in the current iteration to optimize the parameters to be optimized of the pulse high-voltage silicon stack.

[0059] In the process of iteratively optimizing the parameters to be optimized using the multi-objective rime algorithm, if the simulation results of multiple optimization targets of the pulse high-voltage silicon stack show a large decreasing trend, and the positions of the rime particles do not change significantly in their subsequent iterations, it indicates that the rime particles may have fallen into a local search space that simultaneously reduces the results of multiple optimization targets of the pulse high-voltage silicon stack, and it is difficult for the rime particles to jump out of the local search space in subsequent iterations. Therefore, in this embodiment, the adhesion of the rime particles is increased to expand the search range of the rime particles, so that the rime particles can jump out of the local search space in subsequent iterations to avoid subsequent rime particles converging to the local optimal solution;

[0060] On the contrary, in the iterative process of the rime particles, if the simulation results of multiple optimization targets of the pulse high-voltage silicon stack show a large increasing trend, and the positions of the rime particles change significantly in their subsequent iterations, it indicates that the rime particles have entered the local search space containing the pareto optimal solution and may jump out of the local search space in their subsequent iterations. Therefore, in this embodiment, the adhesion of the rime particles is reduced to narrow the search range of the rime particles, so that the rime particles can iterate in the search space where the pareto optimal solution is located to improve the accuracy of subsequent rime particles converging to the pareto optimal solution.

[0061] Therefore, based on the comprehensive simulation change coefficient and the degree of position change, determining the adhesion adjustment value of each rime particle in the current iteration is specifically:

[0062] The expression for the adhesion adjustment value of each rime particle in the current iteration is as follows: ; In the formula, represents the adhesion adjustment value of the ath rime particle in the current iteration; S represents the comprehensive simulation change coefficient in the current iteration; represents the degree of position change of the ath rime particle in the current iteration; norm( ) represents the normalization function, which is used to map the value of to the range (0, 1). This is because the value range of adhesion in the multi-objective rime algorithm is (0, 1); represents a preset constant greater than 0, which is used to prevent the denominator from being 0. The value of is set manually. In this embodiment, the value of

[0063]

[0064] Figure 2

[0065]

[0066] Furthermore, all types of circuit parameters to be optimized in the pulse high-voltage silicon stack are used as the input of the multi-objective rime algorithm. Among them, the adhesion adjustment value of each rime particle in the current iteration is used as the adhesion of the corresponding rime particle in the next iteration of the multi-objective rime algorithm until the iteration is completed, and all types of optimal circuit parameters in the pulse high-voltage silicon stack are obtained.

[0066] The staff selects the optimal circuit parameter combination that meets the national standard GB / T6351-1998 Semiconductor devices Discrete devices Part 2: Rectifier diodes Section 1 Blank detail specification for rectifier diodes (including avalanche rectifier diodes) with a rated rectification current of less than 100A under ambient or case conditions from all types of optimal circuit parameters for subsequent manufacturing of the pulse high-voltage silicon stack.

[0067] It should be noted that: the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the specific embodiments of this specification have been described. Further, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments.

[0069] The above-described embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them; any modification to the technical solutions recorded in the foregoing embodiments, or any equivalent replacement of some of the technical features, does not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and all should be included within the protection scope of the present application.

Claims

1. A method for manufacturing a pulsed high voltage silicon stack, characterized in that: The method comprises the following steps: Obtain various circuit parameters and targets to be optimized of the pulsed high-voltage silicon stack, and use the multi-objective rime algorithm to iteratively optimize the various circuit parameters to be optimized. The specific process is as follows: S1: For the current iteration and all previous iterations, obtain the simulation results of various targets to be optimized of the pulsed high-voltage silicon stack under different circuit parameters to be optimized in each iteration, and update the coordinates of the positions of each rime particle at the end of each iteration; S2: By analyzing the change trend of the simulation results of various targets to be optimized under any type of circuit parameters to be optimized in the current iteration and all previous iterations, the simulation change values ​​of various targets to be optimized in the pulse high-voltage silicon stack under any type of parameters to be optimized in the current iteration are determined, and based on the average distribution of the simulation change values ​​of various targets to be optimized under all types of parameters to be optimized, the simulation change coefficients of various targets to be optimized in the current iteration are determined to determine the comprehensive simulation change coefficient in the current iteration; S3: determining the position change of each rime particle in each iteration by analyzing the distance between the position update coordinates of each rime particle in each iteration and the previous iteration; evaluating the change trend of the position change in the current iteration and all previous iterations, determining the position change trend value of each rime particle in the current iteration, and determining the position change degree of each rime particle in the current iteration in combination with the position change in the current iteration; S4: Based on the comprehensive simulation variation coefficient and the position variation degree, determining the adhesion adjustment value of each rime particle in the current iteration, so as to optimize the parameters to be optimized of the pulsed high-voltage silicon stack.

2. A method for manufacturing a pulsed high voltage silicon stack as claimed in claim 1, characterized in that: The various types of circuit parameters to be optimized and various types of targets to be optimized of the pulse high-voltage silicon stack are obtained, including: all types of circuit parameters to be optimized include at least the reverse internal resistance, the silicon rectifier tube to ground capacitance, the distributed capacitance to the high-voltage end, the number of series tubes and the PN junction reverse barrier capacitance; all types of targets to be optimized include at least the voltage uniformity coefficient, the voltage utilization rate and the reverse withstand voltage value.

3. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The method for determining the simulation change value of each type of target to be optimized in the pulsed high voltage silicon stack under any type of parameter to be optimized in the current iteration is: For the current iteration and all previous iterations, the simulation results of various types of targets to be optimized under any type of circuit parameters to be optimized are fitted to obtain a fitting straight line, and the slope of the fitting straight line is used as the simulation change value of various types of targets to be optimized in the pulse high-voltage silicon stack under any type of circuit parameters to be optimized in the current iteration.

4. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The simulation variation coefficients of various types of targets to be optimized in the current iteration are the average values ​​of the simulation variation coefficients of various types of targets to be optimized in the pulse high-voltage silicon stack under all types of circuit parameters to be optimized in the current iteration.

5. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The expression of the comprehensive simulation variation coefficient under the current iteration is: ; In the formula, Indicates the comprehensive simulation change coefficient under the current iteration; Indicates the number of simulation variation coefficients greater than 0 among the simulation variation coefficients of all targets to be optimized in the current iteration and all previous iterations; It represents the simulation variation coefficient of the nth type of target to be optimized in the pulse high-voltage silicon stack under the current iteration; N represents the total number of categories of targets to be optimized in the high-voltage silicon stack under the current iteration; exp( ) represents an exponential function with a natural constant as the base.

6. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The position change amount of each rime particle in each iteration is the Euclidean distance between the position update coordinates of each rime particle in each iteration and the previous iteration.

7. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The method for determining the position change trend value of each rime particle in the current iteration is: The position change of each rime particle in the current iteration and all previous iterations is used as the ordinate for fitting, and a fitting straight line is obtained and recorded as the position fitting straight line. The slope of the position fitting straight line is used as the position change trend value of each rime particle in the current iteration.

8. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The expression for the position change degree of each rime particle in the current iteration is: ; In the formula, Indicates the degree of position change of the ath rime particle in the current iteration; Indicates the position change of the ath rime particle in the current iteration; It represents the position change trend value of the ath rime particle in the current iteration; exp() represents an exponential function with a natural constant as the base.

9. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The expression of the adhesion adjustment value of each rime particle in the current iteration is: ; In the formula, represents the adhesion adjustment value of the ath rime particle in the current iteration; S represents the comprehensive simulation change coefficient in the current iteration; Indicates the position change degree of the ath rime particle in the current iteration; norm() represents the normalization function; Indicates a preset constant greater than 0.

10. The method for manufacturing a pulsed high voltage silicon stack according to claim 1, characterized in that: The optimizing the parameters to be optimized of the pulsed high voltage silicon stack comprises: All types of circuit parameters to be optimized in the pulse high-voltage silicon stack are used as the input of the multi-objective rime algorithm, wherein the adhesion adjustment value of each rime particle in the current iteration is used as the adhesion of the corresponding rime particle in the next iteration of the multi-objective rime algorithm until the iteration is completed, and the optimal circuit parameters of all types in the pulse high-voltage silicon stack are obtained.

Images